I Thought the ‘Magnificent Seven’ Trade Was Toast Until I Saw What Meta Is Doing With Its AI Chips

Photo of Joey Frenette
By Joey Frenette Published

Quick Read

  • Meta Platforms (META) deployed its custom MTIA 300 inference chip for ranking and recommendation models, with new chip generations arriving every six months to reduce dependence on Nvidia GPUs.

  • Meta is demonstrating early AI returns and building custom chips to compete in inference workloads as the industry shifts from training to deployment, positioning itself to challenge Nvidia’s dominance and compete directly with other hyperscalers.

This post may contain links from our sponsors and affiliates, and Flywheel Publishing may receive compensation for actions taken through them.
I Thought the ‘Magnificent Seven’ Trade Was Toast Until I Saw What Meta Is Doing With Its AI Chips

© nextheprime / Shutterstock.com

The Magnificent Seven trade certainly looked toast going into this year. While the names seemed like excellent long-term ways to profit from the next wave of AI, 2026 looked like it could be a repeat of 2022. While the Mag Seven might take a breather for a while longer as investors come to terms with the CapEx required to play at the AI leader’s table, I do think that some of the names, most notably Meta Platforms (NASDAQ:META | META Price Prediction), are ready to give investors what they want (actual ROI) well ahead of the pack.

The latest quarter already revealed that the AI bets are paying off. And while it’s hardly the payoff that makes up for the magnitude of spend this year, I do think it’s a sign that Meta is, in fact, on the right track and that its AI bets could unlock real value and new market opportunities.

Meta’s AI strategy makes a ton of sense

Personally, I like how Meta is using AI in these early days. Perhaps there’s no better use of AI than to use it to improve one’s own day-to-day operations. Undoubtedly, utilizing AI to improve metrics at one’s own company could be the new scorecard that investors go by. Sure, AI benchmarks and how one model compares to another will always be important to see how the competition stacks up.

However, as AI becomes increasingly specialized, it’ll be more about the metrics and cash flows it can boost. Whether that’s driving some percentage point in engagement or a quantifiable increase to outcomes for customers, I do think that agentic AI is pushing us into a climate where it’s the actual results that rake in the cash flow.

In any case, Meta seems to be doing a lot of things right early on in this AI race. And with the MTIA (Meta Training & Inference Accelerator) chip roadmap recently unveiled, I think it’s time to take the AI, social-media and metaverse titan as a serious player in chips, as it lessens its dependence on the top GPU makers such as Nvidia (NASDAQ:NVDA). Of course, Nvidia GPUs are going to remain in high demand in these earlier days of AI.

However, as the multi-year shift to inference happens, it’s the hyperscalers themselves that could have a chance to really flex their muscles as they look to not only win the AI race, but the AI chip war as well. Apple (NASDAQ:AAPL) knows as well as anyone what’s on the line when it comes to controlling not just software, but the hardware. Given that Meta Platforms is a fierce rival of Apple, it’d only make sense that Mark Zuckerberg’s empire were to follow a similar game plan.

Meta’s custom chip roadmap is seriously impressive

In any case, the new Meta silicon is a beast for AI inference, and with a new chip to come every six months, the social-media turned powerful AI titan might be the hyperscaler with the inference chips to beat in a few years down the road.

Perhaps the six-month cadence could force rivals in the semiconductor scene to accelerate their release schedules as well. Of course, time will tell. Either way, Meta is already eating its own cooking (it’s using its own AI and chips), and if the results are there, that could have customers knocking on its door.

Whether we’re talking about the MTIA 300, which is now in deployment for its own ranking and recommendation models, or the coming MTIA 400/500 chips to come, it seems like Meta has custom silicon down.

As Meta triples down on agents (its Manus and Moltbook deals signify it’s serious about flooring it on agents), it will be interesting to see how the MTIA chips handle Meta’s custom workloads. My guess is that MTIA could be the most efficient hardware to run its own agents. Add continued agentic acquisitions and the super-capable Superintelligence team into the equation, and I think Bill Ackman is right on the money when he says the stock has a “deeply discounted valuation.” 

Even if the Mag Seven stall for the rest of the year, I think Meta is ready to graduate into a class above the group, thanks in part to an AI strategy that seems to already be working.

Photo of Joey Frenette
About the Author Joey Frenette →

Joey is a 24/7 Wall St. contributor and seasoned investment writer whose work can also be found in publications such as The Motley Fool and TipRanks. Holding a B.A.Sc in Computer Engineering from the University of British Columbia (UBC), Joey has leveraged his technical background to provide insightful stock analyses to readers.

Joey's investment philosophy is heavily influenced by Warren Buffett's value investing principles. As a dedicated Buffett disciple, Joey is committed to unearthing value in the tech sector and beyond.

Featured Reads

Our top personal finance-related articles today. Your wallet will thank you later.

Continue Reading

Top Gaining Stocks

CBOE Vol: 1,568,143
PSKY Vol: 12,285,993
STX Vol: 7,378,346
ORCL Vol: 26,317,675
DDOG Vol: 6,247,779

Top Losing Stocks

LKQ
LKQ Vol: 4,367,433
CLX Vol: 13,260,523
SYK Vol: 4,519,455
MHK Vol: 1,859,865
AMGN Vol: 3,818,618